ml4a-guides are a collection of practical resources for working with machine learning software, including code and tutorials.

Fundamentals
Mathematical precepts for machine learning
  • What is machine learning?
  • Vectors, matrices, and tensors
  • Functions, derivatives, and gradient descent
Simple neural networks
How to train a simple feedforward neural network
  • Architecture of simple neural net
  • Training a neural net on toy data
Convolutional neural networks
How to train a convolutional neural network
  • Building neural networks with convolutional and pooling layers for image processing
  • Train a convnet to classify MNIST handwritten digits
Recurrent neural networks
Introduction to recurrent neural nets and LSTMs
  • RNNs, LSTMs, and context
  • Setting up an LSTM to predict characters in text
Sequence to sequence models
Using recurrent neural nets for sequence to sequence models
  • How RNNs operate on sequences on both ends
  • Training LSTM for language translation
Reverse image search
How to do reverse image search, i.e. find similar images to query image
  • Extract feature vectors from images with convnets
  • Find most similar images to query image
Clustering images with t-SNE
How to cluster a collection of images in 2d with a t-SNE
  • Extract feature vectors from images with convnets
  • Embed images in 2d space using a t-SNE over their feature vectors
Clustering sounds with t-SNE
How to cluster a collection of sounds or audio clips in 2d with a t-SNE
  • Extract feature vectors from audio clips with librosa
  • Embed audio clips in 2d with a t-SNE over their feature vectors
Reverse text search
Representing text for search by document
  • Text representation: bag-of-words, tf-idf
  • Latent semantic analysis
  • Search for similar documents
Word vectors
Deriving word vectors/embeddings from a text corpus
  • Embedding words in vector space
  • Geometric operations in word-vector space and analogies
  • Generate a t-SNE from a trained word2vec embedding
Q-Learning
Introduction to Q Learning
  • Introduction to the reinforcement learning problem
  • Implementing a Q-Learner for an agent
Deep Q-Networks
Using deep Q networks for advanced reinforcement learning
  • Deep Q-Networks for larger state spaces
  • Implementing and training DQN for paddle-ball game